On June 11, 2026, U.S. equities rebounded sharply after a weak week, with semiconductor stocks leading the advance. Shares of Penguin Solutions (PENG), onsemi (ON), and Photronics (PLAB) surged by 14%, 9%, and 11% respectively in a single day. Superficially, this rally appears to be a direct reaction to easing inflation expectations and a drop in the 10-year Treasury yield following lower oil prices. But a deeper look reveals that these three companies—representing high-performance computing integration, power semiconductors, and photomask manufacturing—are quietly forming an underappreciated “secondary support triangle” beneath the AI infrastructure boom. In my view, this surge is not merely a rate-driven tactical bounce but a market repricing of the foundational logic underpinning AI’s physical expansion.
The semiconductor sector trades at historically elevated multiples, with a median P/E above 35x, making it acutely sensitive to interest rate shifts. When falling crude prices cooled CPI forecasts, markets swiftly priced in a Fed pause—or even a rate cut—by Q3 2026. While this macro narrative explains broad sector strength, it fails to account for why PENG, ON, and PLAB stood out. Their Q1 2026 earnings reveal that AI-related revenue now accounts for 38%, 27%, and 22% of total sales respectively—well above their five-year averages. More critically, their customer bases significantly overlap: SpaceX uses onsemi’s SiC power modules in Starlink ground stations for energy efficiency; Penguin supplies liquid-cooled HPC clusters to multiple AI training facilities; and Photronics provides critical photomasks for Intel’s 18A node and GAA transistors. This cross-functional synergy forms a hidden yet highly efficient sub-network within the AI hardware supply chain.
Intel plays a pivotal role here. Despite trailing TSMC by roughly 12–18 months in advanced nodes, its IFS (Intel Foundry Services) strategy is transforming it from a pure-play IDM into an open foundry. In May 2026, Intel deepened its collaboration with Photronics to co-develop High-NA EUV photomasks for the 18A node. This not only accelerates Intel’s process roadmap but also opens a new high-end market entry point for Photronics—previously dominated by Japan’s Toppan and firms in Taiwan, China. I judge that part of Photronics’ recent rally reflects growing investor confidence in its technological breakthroughs and tighter integration with Intel’s foundry ambitions.
Penguin Solutions, meanwhile, has long been undervalued. Unlike NVIDIA, it doesn’t sell headline-grabbing chips, but it executes the critical “last mile” of AI deployment: integrating and cooling massive training clusters. Its liquid-cooling systems achieve PUEs below 1.08—crucial for compliance in energy-regulated regions like California. As AI models exceed trillion-parameter scales, cluster sizes balloon, and cooling now accounts for over 18% of total cost of ownership (TCO). Penguin’s value proposition is evolving from engineering services to energy-efficiency infrastructure.
onsemi’s story is more subtle. Often pigeonholed as an automotive chipmaker, it is rapidly monetizing its AI power management portfolio. In Q1 2026, shipments of its intelligent power devices for AI server VRMs (voltage regulator modules) grew 210% year-over-year. Though low in unit price, these components act as the “power gatekeepers” ensuring GPU cluster stability. As AI compute density grows exponentially, power semiconductors are shifting from peripheral to central.
Caution is warranted. Combined, these three companies have a market cap under $50 billion—less than 2% of NVIDIA’s $2.8 trillion valuation. Their rally likely reflects capital rotation away from crowded AI mega-cap trades rather than a fundamental inflection. If strong labor data delays Fed easing, these high-beta, low-liquidity names could quickly give back gains.
Yet if AI infrastructure investment evolves from “chip-centric” spending toward full-stack efficiency—prioritizing thermal management, power integrity, and manufacturing yield—then the secondary layer represented by Penguin, onsemi, and Photronics deserves sustained re-rating. The real test lies ahead: when AI capex shifts from GPU procurement to system-level optimization, will the market finally pay a premium for these silent pillars?